Musk’s rumored $119B Texas “Terafab” could reshape a county, and supercharge his AI ambitions

Elon Musk is eyeing a semiconductor-and-compute megafactory in rural Texas with a price tag that starts at $55 billion and could balloon to $119 billion, numbers that surfaced not in a splashy keynote, but in a local government notice.

The project, dubbed “Terafab,” would rise near the Gibbons Creek Reservoir, about 81 miles northwest of Houston, according to public paperwork from Grimes County. A key next step comes June 3, when local officials hold a public hearing on a property-tax abatement package that could help lure the facility, and lock in what residents get in return.

A county notice puts a dollar figure on Musk’s next big bet

The clearest snapshot of Terafab so far comes from Grimes County, which published a notice describing an initial investment of roughly $55 billion, with total spending potentially reaching $119 billion if later phases move forward.

In Texas, these notices often function as the opening bid in a familiar playbook: local governments offer tax breaks to land major industrial projects, while companies promise jobs, construction, and long-term investment. The scale here is what jumps off the page, this is the kind of project that can remake a county’s tax base, infrastructure priorities, and politics.

But the same document also highlights what’s still fuzzy. The $119 billion figure depends on additional phases that aren’t guaranteed, and the public materials don’t spell out a production start date or a ramp-up timeline. Musk, known for aggressive targets and shifting schedules, hasn’t publicly pinned down when Terafab would actually begin turning out chips.

Why the Gibbons Creek site matters: power, cooling, and local pushback

The proposed location near Gibbons Creek Reservoir isn’t just a dot on a map. The man-made lake was built in the 1980s to help cool a power plant, exactly the kind of infrastructure adjacency that matters for chipmaking and large-scale computing, both of which guzzle electricity and require serious cooling.

Musk has floated a headline-grabbing goal: producing a “terawatt” of computing power per year. A terawatt is one trillion watts, an eye-popping figure that, in public discussions around the project, has been compared to the overall scale of U.S. electricity generation. The comparison isn’t apples-to-apples, but it captures the message Musk wants to send: Terafab would aim for a level of compute output rarely associated with a single site.

That ambition immediately raises a Texas-sized question: where does the power come from? Even in an energy-rich state, adding a massive new industrial load can strain local grids, trigger costly transmission upgrades, and spark community concerns about reliability, especially after Texans lived through high-profile grid failures in recent years.

The customers would be Musk’s own companies, SpaceX, Tesla, and xAI

Terafab is pitched as a pipeline for AI, robotics, and space-related data centers. When Musk discussed the concept earlier this year, he suggested Tesla and SpaceX would jointly steer the effort. But the Grimes County document reportedly references SpaceX specifically, an early detail that could hint at how the project is being structured legally, or how risk is being allocated in the first phase.

The business logic is straightforward: Musk argues Tesla and SpaceX’s appetite for computing power is growing so fast it could outstrip what the global chip market can reliably supply. Building an in-house chip and compute supply chain would be a classic vertical-integration move, less dependence on outside vendors in a world where chip shortages and geopolitical tensions have repeatedly rattled the industry.

Public information around the project suggests chips could feed SpaceX, Musk’s AI venture xAI, and Tesla, supporting everything from model training and automation to communications and data-heavy space ambitions. A built-in customer base would help justify the investment. It wouldn’t make the manufacturing challenge any easier.

Intel and top chip-equipment makers are reportedly in the mix

To pull off a modern chip fab, money is only the entry fee. The reporting around Terafab points to potential involvement from Intel and outreach to Applied Materials, Tokyo Electron, and Lam Research, three of the most important suppliers of the specialized tools used to manufacture semiconductors.

Those machines, used for deposition, etching, cleaning, and metrology, are among the most complex pieces of industrial equipment on Earth, and they often come with long lead times. Even a deep-pocketed buyer has to wait its turn, install the tools, and build a workforce capable of running them at high yield.

Terafab is described as “vertically integrated,” suggesting a site that could span multiple steps, from fabrication to packaging to integration into advanced computing systems. That kind of end-to-end setup can reduce dependencies. It also multiplies execution risk, especially for a newcomer to semiconductor manufacturing.

A moonshot scale meets real-world constraints

The promise is enormous: a factory that doesn’t just make chips, but effectively “manufactures compute” at industrial scale, supporting AI systems on Earth and, eventually, in space. The pitch fits Musk’s broader playbook: build the infrastructure yourself, then use it to accelerate everything else.

The bottlenecks are just as real. Chipmaking demands elite talent, stable supply chains, tight process control, and relentless iteration. And the facility’s success would hinge on basics that don’t bend to hype, power contracts, transmission capacity, cooling, construction timelines, and the ability to consistently produce high-quality chips.

For Grimes County, the June 3 hearing is where the futuristic talk collides with local governance: what tax breaks are on the table, what commitments come back to the community, and how much risk residents are being asked to absorb for a project that, if it materializes, could permanently change the region’s economy and infrastructure.